Cross-linguistic Influences on Infant Vowel Productions Abdulsalam Alhaidary, Karen Mattock, Susan Rvachew, and Linda Polka School of Communication Sciences and Disorders, McGill University, Montreal QC 05 February 2008 I ntroduction  The age at which ambient language begins to shape vowel categories to be language specific is still controversial with some researchers stating that cross-linguistic differences do not emerge until well after the emergence of meaningful speech [1].  Previous studies have reported cross-linguistic variation in the location of the centre of the infant vowel space in the prelinguistic period [2, 3].  Efforts to document cross-linguistic differences in vowel inventories have been hampered by small sample sizes and unreliable transcription of infant speech. Purpose of study  To expand on this earlier work by searching for cross-linguistic differences in infant production of grave corner vowels in a study involving large samples of both infant vocalizations and adult listeners. Hypothesis  Cross-linguistic differences is found in the frequency of occurrence of the vowel /u/ in the grave corner vowel of the infants’ vowel spaces, reflecting differences in the inventory of high vowels found in English versus French.  More grave vowels will be observed in both language groups as infant age increases. Method Participants Infants  Canadian infants aged between 8 and 18 months who were learning either English or French (nCE = 24; nCF = 29). Listeners  5 adult Anglophones and 5 adult Francophones with experience in phonetic transcription. Procedure Vowel selection:  The first formant (F1) and second formant (F2) frequencies of vowels produced within a canonical syllable [1] were obtained.  The diffuse-compact (F2- F1) and grave-acute ([F1+F2]/2) [4] features were calculated for each vowel.  The vowels with most extreme 10% values with respect to these features in each sample were selected, yielding a set of 582 infant vowels. Listener judgments:  Corner vowels [i.e. those with maximum and minimum 10% value of (F1+F2)/2 and (F2- F1)] were submitted for perceptual judgment by native speakers  Listeners choose one out of [i], [æ], [u], [ɑ] or ‘other’ for each vowel. Results  Listeners’ judgments for vowels in the extreme 10% grave corner in the vowel space [minimum (F2- F1)] were submitted for statistical analysis.  The Listener judgment for grave vowel was coded as ‘1’ for expected judgment (i.e. [u]) and ‘0’ for unexpected judgment (i.e. [i], [u], [æ], [ɑ] and ‘other’).  Listener judgments were collapsed across listener groups and each vowel received a score ranging from 0-10. Discussion  Our finding of expansion of the infant vowel space with age towards its corners, in this case grave corner, is in line with previous research. More high- back vowels were perceived as infant age increases [3, 5, 6].  Cross-linguistic differences in infant production of grave corner vowels (i.e. more [u]s in English sample) could be because /u/ appears to be strong attractor for CE infants.  A support for this assumption is stemmed from: (1) Less crowded English vowel space in which any variation on the production of /u/ for CE infants may result in a vowel that may not sound like any English vowel which is not the case for CF infants. (2) Cross-linguistic difference in CE and CF infants’ preference to the corner vowels in which the preference for /u/ versus /y/ is maintained in 12- month old CE infants whereas CF infants show weaker preference for /u/ because both these vowels are phonemic in French [7]. Conclusion  Early changes in infant vowel space are not due solely to maturational changes in oral structure and function; rather the process is more complex and involves the infant’s intake of ambient language input. This study supports the babbling drift hypothesis. Acknowledgments : This research was supported by grants from Natural Sciences and Engineering Research Council of Canada and the Canadian Language and Literacy Research Network to both the third and last authors; post-doctoral fellowship from the Center for Research in Mind, Language and Brain to the second author; and graduate scholarship to the first author from King Saud University, Saudi Arabia. References 1.Oller, D.K., The emergence of the speech capacity. 2000, Mahwah, N.J.: Lawrence Erlbaum Associates. xvii, 428 p. 2.de Boysson-Bardies, B., et al., A crosslinguistic investigation of vowel formants in babbling. Journal of Child Language, 1989. 16: p. 1-17. 3.Rvachew, S., et al., Developmental and cross-linguistic variation in the infant vowel space: the case of Canadian English and Canadian French. J Acoust Soc Am, 2006. 120(4): p. 2250-9. 4.Kuhl, P.K., et al., Cross-language analysis of phonetic units in language addressed to infants. Science, 1997. 277(5326): p. 684-686. 5.Buhr, R.D., The emergence of vowels in an infant. Journal of Speech & Hearing Research, 1980. 23(1): p. 73-94. 6.Kent, R.D. and A.D. Murray, Acoustic features of infant vocalic utterances at 3, 6, and 9 months. J Acoust Soc Am, 1982. 72(2): p. 353-65. 7.Molnar, M. and L. Polka, Vowel preferences in infancy: The role of early language experience, in the Boston Child Language Conference, Boston, Mass. 2004, November: Boston, Mass. 0 1 2 3 4 5 6 7 8 9 10 200 250 300 350 400 450 500 550 600 Age (Days) Number of [u] Judgments French Infants English Infants Linear (French Infants) Linear (English Infants) Figure 2. Number of /u/ judgments for each grave vowel by infant age and language group. One way-ANOVA revealed a significant effect of infant age in which there is an increase of /u/ judgment as the infant age increases [(F [1,144]) =22.67; p<0.0005, η2 = 0.14. In addition, there is a significant effect of infant language in which the frequency of grave vowels judged as [u] by both listener groups was greater among CE infants than among CF infants ((F [1,144]) =7.34; p<0.008, η2 = 0.05). 300 500 700 900 1100 1300 1500 1000 1500 2000 2500 F2 (Mels) F1 (Mels) Diffuse Grave Compact Acute Grave Min (F1+F2)/2 Compact Min (F2-F1) Diffuse Max (F2-F1) Acute Max (F1+F2)/2 [u] [i] [ɑ] [æ] Figure 1. Corner vowels selected from the infants’ speech samples, plotted in F1-F2 space. In the adult system of both language groups, vowels that are in the diffuse, compact, grave and acute corners should sound like [i], [ɑ], [u] and [æ] respectively. Notice that the selected infant corner vowels are overlapping and also some of these vowels may not sound like adult corner vowels.